Refining early Basque criteria (miau)

[Ralf-Stefan Georg]

>Without casting any aspersions on your judgments as a highly competent
>professional linguist, unless you can state "operationally coherent"
>definitions of your prescreening criteria, your choices can look VERY
>arbitrary.  And if you can state operable definitions, the computer should be
>doing the "choosing" to confirm the objectivity of the distinctions being
>made.  That is the "control" that would be expected in other fields when one
>would claim that a computer is confirming ones choices.

Though I'm not sure I'm on the height of the current discussion, I think
this is about excluding descriptive terms from word lists used to "prove"
genetic relatedness (or, for that matter, to form hypotheses about it). I
think, the operational definition Steve Long is asking for can be given
without much ado.

- Exclude all words which arouse the slightest suspicion that any sensual
(mostly, of course, acoustical) perception which might be associated with
the real-world item they denote may have played a role in shaping its
actual form. This rules any kind of imitatives and expressive terms.
Any real language family will be able to stand on its own legs, without
such items taken into account. Any proposal which has to resort to these is
in dire straits anyway.

Of course, with linguae quarum affinitas est demonstrata (as opposed to
demonstranda), it may well be the case that some highly expressive term may
after all turn out to have come into existence due to some regular
processes which derived from some less expressive root/word (as, for
example, it is the case with Russian /zhvachki/ "chewing-gum" - a truly
expressive word for my taste - which remounts to some very unexpressive
root for "chew" in Slavic and, even less so, IE). Again, a successful
demonstration of relatedness will be able to do without them, if the
relationship really holds water.

The reason for the necessity to exclude expressives is not hard to guess -
it is simply that in the *systems* made up by linguistic items (so-called
"languages") the very ability of human vocal tracts to (approximatively)
imitate some noises found in the outside world operates as a strong
attractor. And of course, we know that attractors (strange or not so
strange) disturb systems. But without demonstrating systematicity in
correspondences we have demonstrated nothing. Relationship is demonstrated,
or, better, the demonstration of relatedness is strongest, the more items
are shown to coincide which conform to the good old principle of
"l'arbitraire du signe". Of course we do know, that this principle is not
all-pervasive in language, iow. that iconocity is with us on various
levels, so those items have to go without mercy or remorse. So, the
operational definition is easy: throw the expressives out, here, unlike in
law, a first suspicion dictates the sentence. Every sensible linguist who
is doing classificatory work mainly to advance our knowledge about it will
do this. There are of course others - *certainly* not on this list, but
I've met them on others - who do classification mainly to advance their
careers or their standing in some very exclusive circles, where it is of
prime importance to claim something new or unorthodox - then, of course,
different rules are to be observed. But this need not detain us here.

St.G.

Stefan Georg
Heerstrasse 7
D-53111 Bonn
FRG
+49-228-69-13-32